Co-Targeting of MDM2 and CDK4/6 with Siremadlin and Ribociclib for the Treatment of Patients with Well-Differentiated or Dedifferentiated Liposarcoma: Results from a Proof-of-Concept, Phase Ib Study

This article is featured in Highlights of This Issue, p. 1053

Well-differentiated (WDLPS) and dedifferentiated (DDLPS) liposarcoma are characterized by a consistent co-amplification of murine-double minute-2 (MDM2, up to 100% of cases) and cyclin-dependent kinase-4 (CDK4, ∼90% of cases), located in the 12q13-15 chromosomal region (1–3). Although WDLPS is usually low-grade, nonmetastasizing disease (2, 4), DDLPS is more aggressive and likely to recur and metastasize (5, 6).

Surgery is the cornerstone of treatment for patients with localized disease (2, 4). In the advanced setting, doxorubicin has been the standard treatment of advanced soft-tissue sarcomas for more than 45 years (7), but is associated with very limited efficacy as illustrated by objective response rates between 10% and 23% and a median overall survival of 12 to 22 months (7–10). Novel therapies for advanced WDLPS/DDLPS are therefore urgently needed.

A few studies have investigated agents targeting MDM2 in patients with advanced WDLPS/DDLPS (11, 12). Although some patients showed clinical benefit, this was limited, suggesting that such agents should be assessed through combinatorial regimens (11–13). The combination of cytotoxic drugs, such as doxorubicin, with MDM2 antagonists resulted in a high rate of Grade 3 or 4 hematologic toxicity, with 60% of patients experiencing severe neutropenia and 45% experiencing thrombocytopenia, precluding future development (14). Therefore, combining MDM2 antagonist with a targeted drug may represent a more relevant approach. CDK4 antagonists have shown some clinical activity in patients with advanced WDLPS/DDLPS (15–17). Preclinical data have suggested that CDK4 inhibitors could synergize with MDM2 antagonists in WDLPS/DDLPS models in vitro and in vivo (13). Siremadlin (an inhibitor of the p53–MDM2 interaction) and ribociclib (a CDK4/6 inhibitor) have been shown to be safe and displayed early signs of clinical activity as single agents in patients with advanced cancer (15, 18, 19). We hypothesized that combining siremadlin with ribociclib could activate two critical tumor suppressor pathways and promote anticancer activity in patients with advanced WDLPS/DDLPS. The primary objective of this proof-of-concept, phase Ib study was to determine the regimen, maximum tolerated dose (MTD), and recommended dose for expansion (RDE) for this combination in this specific population.

All animal experiments were performed according to procedures covered by permit number BS-1975 issued by the Cantonal Veterinary Office, Basel, Switzerland, and strictly adhered to the federal animal protection act and the federal animal protection code. All animals were permitted to adapt for 7 days and housed in a pathogen-controlled environment (5 mice/Type III cage) with access to food and water ad libitum and were identified with transponders. LP6 is part of the Cancer Cell Line Encyclopedia. Copy number for MDM2 and CDK4 were analyzed by quantitative polymerase chain reaction at the time of the trial and LP6 cells were confirmed by single-nucleotide polymorphism profiling. After 9 days, the mice were randomized (n = 6) and treatment was initiated. Weekly treatment of siremadlin in mice was composed of three injections every 3 hours on the day of treatment to compensate for the highest clearance in this species. The combination index was adapted from the Clarke definition (1), with values below 0.7 indicating synergism.

This was a proof-of-concept, phase Ib, multicenter, open-label, dose-escalation study of treatment with siremadlin and ribociclib in patients with advanced WDLPS or DDLPS that had radiologically progressed on, or despite, prior systemic therapy. Eligible patients were aged 18 years or older, with Eastern Cooperative Oncology Group (ECOG) performance status ≤1. Patients must have had locally advanced or metastatic WDLPS/DDLPS, received at least one prior systemic therapy, and had radiographic progression per Response Evaluation Criteria in Solid Tumors (RECIST) v1.1 during or within 6 months of their last systemic therapy prior to enrollment. Patients had mandatory biopsies at screening and during treatment. Patients were excluded if they had received prior MDM2 or CDK4/6 inhibitors or had known TP53 mutant tumors. Additional exclusion criteria included symptomatic central nervous system metastases requiring increasing doses of steroids or local central nervous system–directed therapy, concurrent other malignancy, clinically significant uncontrolled heart disease or history of cerebrovascular events within the last 3 months, impairment of gastrointestinal function that may affect absorption of the study drugs, previous therapy, including prior radiation therapy that included >30% of bone marrow reserve within 6 months of treatment start and chemotherapy, biologic therapy, prior investigational study, or major surgery within 2 weeks of treatment start.

Written informed consent was obtained from each patient. The study adhered to the ethical principles of the Declaration of Helsinki and complied with Good Clinical Practice. The protocol was reviewed and approved by the Independent Ethics Committee or Institutional Review Board for each center.

Two regimens were initially explored: Regimen A [4-week cycle: siremadlin and ribociclib once daily (QD; 2 weeks on, 2 weeks off)] and Regimen B [3-week cycle: siremadlin once every 3 weeks (Q3W); ribociclib QD (2 weeks on, 1 week off)]. In addition, based on emerging safety and pharmacokinetic data, Regimen C [4-week cycle: siremadlin once every 4 weeks; ribociclib QD (2 weeks on, 2 weeks off)] was tested as an alternative to Regimen B to mitigate the risk of potential bone marrow toxicities. All dose levels tested are presented in Supplementary Table S1. Patients continued treatment until unacceptable toxicity, disease progression, and/or treatment discontinuation at the discretion of the investigator or patient. Up to three dose reductions were allowed per patient, after which treatment was discontinued. For analyses, data from Regimens B and C (in which siremadlin was administered as a pulsed, high dose) were combined.

The Full Analysis Set comprised all patients who received at least one dose of study treatment and was the analysis set used for all analyses unless specified otherwise. The Safety Set included all patients who received at least one dose of study treatment and had at least one valid post-baseline safety assessment and was the analysis set used to assess all safety endpoints except the dose-dose-limiting toxicity (dose-DLT) relationship. The Pharmacokinetic Analysis Set consisted of all patients who had at least one blood sample providing evaluable pharmacokinetic data and was used to summarize all pharmacokinetic data and derived pharmacokinetic parameters. Statistical analyses of biomarker data were considered exploratory in nature. The Dose-determining Set consisted of all patients enrolled in the phase Ib part and included in the safety set who either met the minimum exposure criterion and had scheduled safety evaluations or experienced a DLT during the first cycle of treatment. A patient was considered to have met the minimum exposure criterion if the patient had received at least 75% of planned doses of siremadlin and 75% of planned doses of ribociclib in the first cycle of dosing.

Safety assessments were performed on the Safety Set except for assessment of dose-DLT relationship, which was performed on the Dose-determining Set. Safety was monitored using physical examination, vital signs, weight, ECOG performance status, laboratory evaluations, and pregnancy and cardiac assessments. Adverse events (AEs) were recorded at every visit, and toxicity was assessed using the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) v4.03 and Medical Dictionary for Regulatory Activities (MedDRA) v21. An AE of special interest (AESI) is a grouping of AEs that are of scientific and medical concern specific to the treatment under investigation in this study. A DLT was defined as an AE or abnormal laboratory value considered unrelated to the disease, disease progression, intercurrent illness, or concomitant medications, which occurred within the first cycle (21 days for Regimen B; 28 days for Regimens A and C) of treatment.

Efficacy assessments were performed on the Full Analysis Set. Tumor response was assessed locally by investigators based on RECIST v1.1. The local investigator's assessment was used for the efficacy analysis and treatment decision-making. CT or MRI was performed at baseline within 28 days of treatment start. During treatment, tumors were assessed every 6 weeks for 6 months and every 12 weeks thereafter until disease progression.

Pharmacokinetic analyses were performed on the Pharmacokinetic Analysis Set. Depending on the regimen, serial blood samples were collected from all patients at specified time points on Days 1 and 14 of Cycles 1 and 2, to assess single- and multiple-dose pharmacokinetic properties of siremadlin and ribociclib. Noncompartmental pharmacokinetic parameters were estimated from individual plasma concentration–time profiles and included the area under the concentration-time curve from the time of dosing to the last measurable concentration (AUC_last), the AUC from administration to last observed concentration at time t (AUC_0-t), the maximum observed concentration (C_max), and the median time to reach maximum plasma concentrations (T_max). Plasma concentrations were determined using a validated liquid chromatography method with tandem mass spectrometry and with 1.0 ng/mL as the lower limit of quantification. Pharmacokinetic parameters were estimated through noncompartmental methods using version 6 of Phoenix software (Pharsight Corporation), with actual post-dose time used for the estimation of pharmacokinetic parameters instead of nominal time.

Biomarker assessments were performed on tumor and surrogate tissues (blood) collected before and during treatment. In Regimen A, tumor samples were taken at screening and Cycle 1, Day 14 (−6 days), and blood samples throughout Cycle 1, pre- and post-dose. In Regimens B and C, tumor samples were taken at screening and Cycle 2, Day 2 (or Cycle 2, Day 3), and blood samples throughout Cycle 1 and Cycle 2, pre- and post-dose. Growth differentiation factor-15 (GDF-15) in serum was measured using enzyme-linked immunosorbent assay (2). Genetic alterations in pretreatment tumors were assessed using next-generation sequencing and the FoundationOne test. A patient was determined to be un-evaluable if there were no known or likely copy-number alterations (CNAs), rearrangements, or short variants of functional significance identified.

The primary objective was to determine the MTD and/or RDE of the combination in one or more of three predefined regimens by investigating the incidence of DLTs during the first cycle of treatment and exposure to siremadlin and ribociclib. Secondary objectives included characterizing safety and tolerability, pharmacokinetic and pharmacodynamic properties, and preliminary antitumor activity.

Best overall response was the best response recorded from treatment start until disease progression or recurrence. Overall response rate was the proportion of patients with a best overall response of complete response or partial response. Disease control rate was the proportion of patients with a best overall response of complete response, partial response, or stable disease lasting 23 weeks or longer. Certain subtypes of LPS progress slowly, and stable disease lasting less than 23 weeks may not reflect a treatment effect; median progression-free survival (PFS) lasting 23 weeks or longer was considered as clinically relevant, as it implied a 50% increase in median PFS from a benchmark value for single-agent CDK4/6 inhibitor (16). PFS is defined as the time from treatment start to the first radiologically documented disease progression or death from any cause. If a patient had not had an event at the time of last patient last visit, PFS was censored at the date of last adequate tumor evaluation. Clinical deterioration without objective radiologic evidence was not considered as documented disease progression.

An adaptive Bayesian logistic regression model (BLRM) using the Escalation With Overdose Control (EWOC) principle was estimated to guide dose escalation (20). MTD was estimated using the BLRM and based upon the probability of a DLT occurring during the first treatment cycle for patients in the Dose-determining Set. The MTD was defined as the highest combination of drug doses expected to cause DLT in less than 35% of the treated patients in the first cycle of treatment during the escalation part of the study. After each cohort, a 5-parameter BLRM for combination treatment was fitted on the Cycle 1 DLT data (i.e., absence or presence of DLT) accumulated throughout dose escalation to model the dose-DLT relationship of siremadlin and ribociclib given as combination. Dose recommendation was based on the probability that the true DLT rate for each dose lies in one of the following categories: under-dosing: (0–<16%), targeted toxicity (≥16–<35%), or excessive toxicity (≥35–100%). Following the principle of EWOC, after each cohort of patients, the recommended dose combination for the next cohort had to fulfill the overdose criterion that there is less than 25% chance of excessive toxicity. All information that was available about the dose-DLT relationships of single agents siremadlin and ribociclib was summarized in prior distributions for the model parameters. A weakly informative prior distribution was used for the interaction parameter between siremadlin and ribociclib (21).

Data describing demographic and baseline characteristics, measurements of efficacy and safety, and pharmacokinetic and pharmacodynamic properties were summarized with descriptive statistics. A Wilcoxon rank-sum test was used to determine differential genomic alterations [including amplifications, deletions, and single nucleotide variants (SNVs)] between clinical outcome groups.

This trial is registered with ClinicalTrials.gov, number NCT02343172.

Novartis will not provide access to patient-level data, if there is a reasonable likelihood that individual patients could be re-identified. Phase I studies, by their nature, present a high risk of patient re-identification; therefore, patient individual results for phase I studies cannot be shared. In addition, clinical data, in some cases, have been collected subject to contractual or consent provisions that prohibit transfer to third parties. Such restrictions may preclude granting access under these provisions. Where co-development agreements or other legal restrictions prevent companies from sharing particular data, companies will work with qualified requestors to provide summary information where possible.

In preclinical studies evaluating both daily, low-dose, and pulsed high-dose siremadlin, the combination of siremadlin and ribociclib led to synergistic antitumor activity in the MDM2 and CDK4-amplified DDLPS in vivo model, providing evidence that combined targeting may offer a promising therapeutic strategy for liposarcoma (Supplementary Fig. S1).

Between March 13, 2015 (first patient first visit) and October 16, 2019 (last patient last visit), 74 patients were enrolled into Regimen A (n = 26), Regimen B (n = 29), or Regimen C (n = 19). All patients discontinued treatment; this was due to disease progression (n = 52), AEs (n = 14), physician/patient decision (n = 6), or death (n = 2).

Baseline disease characteristics were generally well balanced across the three regimens (Table 1). The distribution of patients with DDLPS (68%) or WDLPS (15%) was similar to observations in clinical practice (4).

Median exposure was 13 weeks (range, 1–174), with 27% (20/74) of patients exposed to treatment for longer than 24 weeks. Three patients receiving Regimen A, 3 patients receiving Regimen B, and 2 patients receiving Regimen C were exposed to treatment for longer than 52 weeks. Delays to doses of siremadlin (66%) and ribociclib (73%) were comparable overall, although more frequent among patients receiving Regimen B (79% and 86% for siremadlin and ribociclib, respectively), and primarily caused by AEs.

DLTs were reported in 2 patients receiving Regimen A (9%), 9 receiving Regimen B (32%), and 5 receiving Regimen C (28%) (Table 2). All DLTs were Grade 3/4 (except for one Grade 2 event each of neutropenia and prolonged QT interval) and were generally hematotoxic. Regimen A was terminated due to a high incidence of severe hematotoxicities during and beyond the DLT period, together with the absence of clinical activity at the different dose levels tested. Thus, no MTD or RDE was declared for this regimen. Although predefined as an alternative to Regimen B to mitigate expected hematotoxicities, improved tolerance was not observed with Regimen C despite the longer cycle duration. Thus, no MTD or RDE was declared for this regimen either. RDE was based on the BLRM results and a synthesis of all relevant safety, efficacy, pharmacokinetic, and pharmacodynamic data. From these data, the combination of siremadlin 120 mg Q3W and ribociclib 200 mg QD, 2 weeks on, 1 week off (Regimen B), was declared as the RDE. Translated average siremadlin plasma concentrations in human required for tumor stasis in liposarcoma [19 ng/mL, derived from pharmacokinetic/pharmacodynamic modeling of data from liposarcoma patient-derived xenograft rats treated with siremadlin (data not shown)] were generally achieved in the majority of patients starting from the siremadlin dose of ∼100 and ∼130 mg in Regimen B and Regimen C, respectively. Predicted efficacious preclinical concentrations were in line with observed clinical activity.

Most patients (96%) experienced one or more AEs suspected of being treatment-related (Table 2). Gastrointestinal and hematologic toxicities were the most common, although gastrointestinal toxicities were predominantly low grade and not dose-limiting. The three most common treatment-related, nonhematologic AEs (any grade) were nausea (67%), fatigue (54%), and vomiting (50%). The most common (occurring in ≥25% patients) Grade 3/4 treatment-related AEs were decreased platelet count and decreased white blood cell count (each 27%), and neutropenia (25%). All patients experienced one or more AEs, regardless of causality (Supplementary Table S2). Four deaths were reported during or within 30 days of treatment; 1 patient receiving Regimen A and 2 receiving Regimen B died as a result of disease progression, while 1 patient receiving Regimen C (150 mg siremadlin and 300 mg ribociclib) died as a result of hematotoxicity. Febrile neutropenia, pneumonia, fistula, sepsis, neutropenia, and decreased platelet count were ongoing at the time of death, all of which were suspected to be treatment-related. Safety profile with regards to bone marrow was tolerable, even on sustained treatment. Overall, AEs were manageable, patients received supportive care with colony-stimulating growth factors or transfusions as necessary, and more than 90% of dose delays and reductions were attributable to AEs. Regarding serious AEs (Supplementary Tables S3 and S4) and AESIs (Supplementary Table S5), no differences or significant trends were observed between the regimens.

No complete responses were achieved during the study (Table 3; Fig. 1), and partial responses were reported in 3 patients (2 patients in regimen B and 1 in regimen C; Fig. 1); CT images of 1 patient achieving a partial response under Regimen B are shown in Supplementary Fig. S2. Stable disease was achieved by 38 patients in total: 11 patients (42%) receiving Regimen A, 17 (59%) receiving Regimen B, and 10 (53%) receiving Regimen C. Seven patients (27%) receiving Regimen A, 6 (21%) receiving Regimen B, and 4 (21%) receiving Regimen C achieved stable disease lasting 23 weeks or longer (rationale for the choice of the 23-week cut-off is presented in the Materials and Methods section). The disease control rate was 27% [95% confidence interval (CI), 12–48] in Regimen A, 28% (95% CI, 13–47) in Regimen B, and 26% (95% CI, 9–51) in Regimen C. Best overall response summary tables by dose level and regimen are shown in Supplementary Table S6.

Median PFS was higher in patients receiving Regimen B (4.2 months; 95% CI, 2.8–16.5) and Regimen C (4.0 months; 95% CI, 1.4–8.0) compared with patients receiving Regimen A (2.7 months; 95% CI, 1.9–8.2; Supplementary Table S7).

Linear pharmacokinetic characteristics of orally administered siremadlin, as demonstrated over a broad clinical dose range (1–350 mg) with no apparent absorption limitation, was previously described by a 1-compartment linear model (22). Pharmacokinetic parameters from noncompartmental analysis were summarized by dose regimen for siremadlin (Supplementary Table S8) and ribociclib (Supplementary Table S9). Following oral administration of siremadlin, T_max generally ranged from 2 to 8 hours, independent of dose. The interpatient variabilities for AUC–time curve from time 0 to 24 hours (AUC_24h) or to 48 hours (AUC_48h) and C_max were generally moderate. Under Regimen A (in terms of C_max and AUC_24h), mean exposures to siremadlin at Day 14 in the presence of ribociclib increased with increasing siremadlin dose from 5 to 15 mg in an approximately dose-proportional manner (Supplementary Table S8). Exposures (AUC_24h) to siremadlin at Day 14 in the presence of 400 mg ribociclib were, in general, 2- to 2.5-fold higher than those observed for the single-agent alone (23); this is consistent with a preclinical predicted drug–drug interaction (from physiologically based pharmacokinetic modeling using SimCyp®) showing an increase of siremadlin exposure under Regimen A due to inhibition of its metabolism by ribociclib via hepatic CYP3A4. Under Regimens B and C, exposures (AUC_48h) to siremadlin in combination with ribociclib were generally within the range observed for the single-agent alone. Siremadlin did not appear to alter the pharmacokinetic properties of ribociclib (Supplementary Table S9; ref. 15).

Pharmacodynamic changes in direct transcriptional target of p53 (GDF-15) were used to measure siremadlin-driven p53 pathway activation. The GDF-15 promoter region contains p53 response elements that confer p53-specific transactivation and its subsequent gene activation (24), which is dependent on wild-type p53 (25). GDF-15 induction and secretion can be measured in blood, which has been demonstrated to directly correlate with p53 pathway activation (26). As a surrogate for target inhibition, modulation of GDF-15 was measured in siremadlin pre- and post-dose serum samples. GDF-15 levels, depicted as a fold change from baseline, increased with increasing dose levels of siremadlin, demonstrating that clinically relevant drug exposures were achieved and modulation of downstream p53 targets was accomplished (Fig. 2). To determine how baseline biomarker status may relate to clinical outcome, next-generation DNA sequencing was performed on pretreatment tumor samples. Detection of gene mutations [substitutions, insertion and deletion alterations (indels), and CNAs] in 324 genes were evaluated using the FoundationOne panel. The presence of MDM2/CDK4 co-amplification was detected in 83% (33/40) of patients, which is in line with published reports (27, 28). Detected pretreatment somatic alterations of known and likely significance are presented in Fig. 3A. Best overall response, as well as best percent change in tumor size, are also depicted. Specific somatic alterations that correlate with clinical outcome, assessed using chi-squared tests for each gene, could not be identified. Notably, the 2 patients who achieved partial response presented with fewer somatic genetic alterations compared with the rest of the patients. Aside from near obligate amplification in MDM2, CDK4, and/or FRS2, also observed in most of the patients on the study and associated with the disease (29), no other gene alterations were detected in patients achieving partial response (Fig. 3A). In addition, the total number of genomic alterations (amplifications, deletions, and SNVs) in patients who experienced some tumor shrinkage or disease control longer than 23 weeks was significantly lower than in those patients who progressed quickly (P = 0.00335; Fig. 3B and C).

This represents the first proof-of-concept study investigating co-targeting of MDM2 and CDK4, a key genetic feature of WDLPS and DDLPS (30–32). This is also one of the largest trials in liposarcoma reported to date, enrolling patients with locally advanced or metastatic disease and unequivocal radiologic progression, and which systematically collected on-treatment biopsies.

The combination of MDM2 antagonists with cytotoxic drugs such as doxorubicin resulted in a high rate of Grade 3 or 4 hematological toxicity, with 60% of patients experiencing severe neutropenia and 45% experiencing thrombocytopenia, precluding future development (14). Therefore, combining an MDM2 antagonist with a targeted, less cytotoxic drug may represent a more relevant approach.

The combination demonstrated a manageable safety profile that was no less tolerable than those of the single agents (15, 18, 19). Furthermore, safety results were consistent with other agents used to treat liposarcoma such as the microtubule-dynamics inhibitor, eribulin (33). Sixteen patients experienced mostly hematotoxic DLTs, including neutropenia and thrombocytopenia. The RDE was established as siremadlin 120 mg Q3W and ribociclib 200 mg QD, 2 weeks on, 1 week off (Regimen B). Regimen C was explored to evaluate whether a longer time off treatment could mitigate safety risk while maintaining clinical activity; however, this was not demonstrated.

The disease control rate was 28% in patients receiving Regimen B; 2 patients achieved a partial response, 6 achieved stable disease lasting 23 weeks or longer, and median PFS across all doses tested under Regimen B was 4.2 months. This preliminary activity was encouraging considering the sample size. The 3-month PFS rates of 43.8% under Regimen A, 65.9% under Regimen B, and 55.6% under Regimen C were all higher than the 40% threshold for considering a drug active in patients with advanced soft-tissue sarcomas (34). Importantly, progression according to RECIST v1.1 within a 6-month interval was mandatory for inclusion in this study, allowing a more direct measure of antitumor activity attributable to treatment.

Our study demonstrates evidence of target engagement through pharmacodynamic modulation of direct transcriptional targets of p53. Promising results were observed for GDF-15, which was induced dose-dependently by siremadlin, confirming that clinically relevant drug exposure and modulation of downstream p53 targets was accomplished. Pharmacokinetic data for siremadlin in combination with ribociclib were generally consistent with single-agent siremadlin pharmacokinetic properties, albeit with mean exposures trending slightly higher versus the single agent, suggesting ribociclib has a mild impact on the pharmacokinetics of siremadlin in the regimen selected as the RDE.

The combination of siremadlin and ribociclib demonstrated a manageable safety profile, with no drug–drug interactions in terms of pharmacokinetics in the regimens selected as the RDE. Even though encouraging preliminary activity was reported in these patients with advanced liposarcoma, the combination of MDM2 and CDK4 inhibitors was expected to be highly synergistic, based upon preclinical experiments (Supplementary Materials and Methods). However, expected synergism preclinically did not translate into patients in the clinic, with regards to the regimens tested in this trial. Several factors may have contributed to this: the novel combination was tested across various doses and regimens in dose escalation, and enrolled patients presented heterogeneous characteristics, including variable anatomical locations and a wide range of prior treatment lines.

Chemotherapy may have selective pressure on p53-mutant clones, which were not captured in pretreatment biopsies (35). As noted in this study, WDLPS/DDLPS frequently exhibit heterogeneous areas of differentiation and may have diverging results (Fig. 3A). Notably, next-generation sequencing of tumors in 3 patients detected p53 alterations, a negative predictor of response to MDM2 inhibitors (35, 36). All 3 patients were lacking MDM2 amplification and 2 had a concomitant deletion of the RB1 gene, which is a very strong negative predictor of response to CDK4 inhibitors (37, 38). All 3 patients immediately progressed on treatment. The genetic landscape identified in tumors of these patients would suggest that they are unlikely DDLPS. Notably, 2 additional patients with CHEK2 mutations quickly progressed. The role of CHEK2 has not been extensively studied in liposarcoma, but mutations in CHEK2 may affect MDM2–p53 interaction (39). Furthermore, the 2 patients who achieved partial response had fewer somatic genetic alterations compared with the rest of the patients (Fig. 3A). One of the 2 patients with partial response had no other gene alterations aside from MDM2 amplification, further supporting the notion of a better clinical outcome in patients with fewer additional oncogenic alterations.

High-level MDM2 amplification is disease-defining in both WDLPS and DDLPS (1, 2, 4). This is associated with effective suppression of p53, and it may be reasonable to assume that mutations of p53 would be unlikely to provide a growth benefit in the absence of selection pressure with treatment. Hence, studying combined MDM2/CDK4 inhibition in the first-line setting may reveal more clinical activity. Incidence of p53 mutations in relation to the number and types of chemotherapeutic drugs as well as level of differentiation (WDLPS vs. DDLPS) has not been studied yet. In this context, plasma sequencing at baseline may also prove helpful to exclude patients from MDM2-directed inhibitors but was unfortunately not available in this trial.

This study provides insights into the optimal dosing regimen of MDM2 and CDK4 inhibitor combinations. In this study, low-dose daily Regimen A was less efficacious than high-dose, pulsed Regimens B and C consistent with preclinical data (23, 40). In vivo studies with siremadlin indicated that high-dose, pulsed regimens caused rapid and dramatic induction of p53-dependent PUMA expression and apoptosis, versus a daily low dose (23). Our results suggest that induction of apoptosis may be important for the efficacy of siremadlin and ribociclib in liposarcoma, whereas cell-cycle arrest is triggered by the addition of ribociclib. Thus, liposarcoma tumors may need both agents to respond.

The early antitumor activity and manageable safety profile demonstrated by the combination of siremadlin, an MDM2 inhibitor, with ribociclib, a CDK4/6 inhibitor, in this phase Ib study of patients with advanced WDLPS or DDLPS, suggest that this combination may be a useful treatment option in this clinical setting.

A.R. Abdul Razak reports grants from Merck, Bristol Myers Squibb, Novartis, Karyopharm, Boston Biochemical, Deciphera, Genentech, Roche, Pfizer, Medimmune, Eli Lilly, Boehringer Ingleheim, Entremed/CASI Pharmaceuticals, Amgen, Champions Oncology, Iterion, Blueprint, Adaptimmune, and GSK during the conduct of the study, as well as personal fees from Eli Lilly, Boehringer Ingleheim, Merck, Adaptimmune, and GSK outside the submitted work. S. Bauer reports grants and personal fees from Novartis, as well as personal fees from Boehringer Ingelheim and Daiichi Sankyo during the conduct of the study. S. Bauer also reports grants and personal fees from Blueprint Medicines; grants from Incyte; personal fees from Lilly, Adcendio, Deciphera, GSK, Exelixis, and Roche; and personal fees and other support from Bayer outside the submitted work. C. Suarez reports grants and other support from Astellas Pharma, Bayer, Bristol Myers Squibb (Inst), EUSA Pharma, Novartis, Pfizer, S.L.U, Sanofi Aventis, and F. Hoffman-La Roche; other support from Ipsen and Merck Sharp & Dohme; and grants from AB Science, Aragon Pharmaceuticals, AstraZeneca, AB, Blueprint Medicines Corporation, Boehringer Ingelheim España, S.A, Clovis Oncology, Exelixis INC. F., Genentech Inc., and Glaxosmithkline, S.A outside the submitted work. C.-C. Lin reports personal fees from AbbVie, Bayer, BeiGene, Blueprint Medicines, Boehringer Ingelheim, Bristol Myers Squibb, Daiichi Sankyo, Eli Lilly, Merck KGaA, Novartis, and Roche outside the submitted work. R. Quek reports personal fees and other support from BMS; other support from Merck and DKSH; and personal fees from Novartis, Bayer, AztraZeneca, Eisai, and Eli Lilly outside the submitted work. S. Ferretti reports other support from Novartis during the conduct of the study, as well as other support from Novartis outside the submitted work; S. Ferretti also has a patent for WO2015/097622 issued. A. Jullion reports other support from Novartis outside the submitted work. E.J. Orlando reports employment with Novartis. G. Clementi reports personal fees from Novartis during the conduct of the study. E. Halilovic reports a patent for US10220038B2 issued to NOVARTIS AG. C. Fabre reports other support from Novartis during the conduct of the study, as well as other support from Novartis outside the submitted work. J.-Y. Blay reports grants and other support from Novartis and other support from Amgen during the conduct of the study. A. Italiano reports grants from Bayer, Roche, AstraZeneca, BMS, and MSD, as well as personal fees from Bayer and Springworks outside the submitted work. No disclosures were reported by the other authors.

A.R. Abdul Razak: Resources, investigation. S. Bauer: Resources, investigation. C. Suarez: Resources, investigation. C.-C. Lin: Resources, investigation. R. Quek: Resources, investigation. M.L. Hütter-Krönke: Resources, investigation. R. Cubedo: Resources, investigation. S. Ferretti: Resources, investigation. N. Guerreiro: Formal analysis. A. Jullion: Formal analysis. E.J. Orlando: Formal analysis. G. Clementi: Data curation. J. Sand Dejmek: Supervision. E. Halilovic: Formal analysis. C. Fabre: Conceptualization, supervision. J.-Y. Blay: Resources, investigation. A. Italiano: Resources, investigation.

The authors would like to thank Maria Santos Rosa, Novartis, for conceptualization, data curation, formal analysis, and methodology guidance. The authors would also like to thank Wei He, Novartis, for providing continuing statistical support for the study. Medical writing assistance was provided by Ailsa Bennett, PhD, of SciMentum, and was funded by Novartis Pharmaceuticals Corporation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.